The effect of manipulating task constraints on game performance in youth field hockey

Abstract

The purpose of this study was to examine the influence of manipulating game constraints on match performance in youth field hockey. A total of 25 participants aged 10.6–14.6 years old played four different 25-min games where density (228 m² or 158 m² per player) and/or number of players (11 per side or 8 per side) was manipulated. Match performance was determined by using notational analysis and physical demands were determined by using GPS analyses. Manipulating the number of players led to an increase in a successful passes (+2.68 passes), skilled actions (+3.73 skilled actions) and successful actions (+3.77 successful actions) performed per player and also created a more advantageous environment to enhance decision making. Increasing the density led to a decrease in unsuccessful dribbles (−0.59 unsuccessful dribbles) made by players and an increase in high intensity running (+38 m) and sprinting (+21.2 m). The findings of this study provide an insight into the effect of manipulating task constraints in skilled junior field hockey and the findings highlight that all types of constraints influence emergent performance in their unique way and that coaches should consider these interactions to promote specific playing behaviour.

Keywords

Notational analysis skill small-sided games

Introduction

Sporting organizations around the world are scaling equipment and playing area to create appropriate competition circumstances for young athletes.¹ A widely used method to create more appropriate training environments are small-sided games (SSG), where the number of players and playing area are scaled to influence on the technical and physiological demands of the game.^2–4 It has been demonstrated that increasing the number of players in basketball and soccer led to an increase in the percentage of successful passes and a decrease in the number of passes and dribbles.^5,3 Whereas similar work in water polo revealed that an increase in number of players led to more passes and offensive actions,⁵ whereas in Australian football manipulating, the numbers of players did not have any effect on technical actions.⁴ Contradictory results can also be found for the effect of playing area on technical playing behaviour when sports such as rugby league, soccer and water polo are compared.^5–7 Indeed, the game dynamics of the different sports vary to a large extent which in part explains the contradictory findings between sports. However, there is no clear evidence for the sport specific differences and the underlying mechanism remains poorly understood.

A possible explanation for this poorly understood mechanism could be related to the mainly descriptive analysis in previous studies.^2–7 Although these studies clearly report about the influence of several SSG manipulations on the frequency of sport-specific technical skills and physiological demands of the SSG, these studies lack a theoretical perspective to explain the changes in playing behaviour in different sports. A useful theoretical framework that may assist to explain differences in SSG research is the constraints-led approach. Newell⁸ emphasized that three different constraints (personal, environmental and task) can all be independently manipulated to guide the acquisition of skill.⁹ Task constraints refer to the characteristics of a specific game such as rules, equipment and playing surfaces. Personal constraints refer to the action capabilities and cognitive capabilities of an individual and the environmental constraints refer to characteristics of the surroundings such as altitude, noise and lighting.

An important feature of the constraints-led approach is the coupling between perception and action.¹⁰ Gibson¹¹ argued that people are surrounded by energy flows which support decision-making and planning of movement. These energy flows contain specific information (affordances) about the environment that can then be used to reach a certain goal. Individuals can pick up these specific affordances and create specific information-movement couplings by becoming attuned to them. Two different processes are suggested to create these specific information-movement couplings.¹² First individuals have to detect the critical information from all the stimuli that is available in the environment and then individuals have to calibrate their actions to this critical information. In SSG, different information is available compared to the adult game. For example, when field dimensions are manipulated, the relative personal playing area changes and can influence the amount of pressure players experience through a reduction in the time available to make a decision. Information-movement couplings may also be influenced by the amount of potential connections (passes) available based on the amount of players per side. An increase in number of players could increase the task complexity due to the increase in potential connections. Both factors can influence the available critical information that players can pick up to create a stable information-movement coupling and perform a successful action.

While much has been learned from previous studies that have examined task constraints such as playing numbers and/or density on game dynamics in team-sports, several gaps in the literature remain. The aim of this study was to examine the impact of manipulating the number of players and relative playing density on match performance in junior field hockey. It is suggested that the two manipulations will influence match performance in different ways because of their unique interaction with the environment.

Methods

Participants

Twenty-five skilled junior field hockey players with an average age of 12.2 ± 0.9 years and a standing height of 1.56 ± 0.11 meters volunteered to participate in this study, after ethical approval was granted by the university ethics committee and parental consent was obtained. All children played in a regional team from their city. These teams play in the zone challenges, a competition organized by the State Hockey association for talented club players from the State.

Experimental design

The number of players and playing density was systematically manipulated during field hockey games. Density or the individual playing area per player in m² was manipulated by comparing the adult game of hockey (228 m² – labelled standard density) with a half-field game (158 m² – labelled scaled density) commonly used in competition for U10 s. The number of players was then co-varied with density with the adult game (11 players a side – labelled standard numbers) compared with eight players per team (labelled scaled numbers) which is a typical playing number in junior competition (Table 1). Teams completed a 25-min field hockey match in each condition and there were 11 participants who played in all four game conditions.

Table 1.

Characteristics of the four different experimental conditions.

	Number of players	Density (m²)	Length/width ratio	Pitch dimensions (m)
Standard density – standard numbers	11 per side	228	±1.6	91 × 55
Scaled density – standard numbers	11 per side	158	±1.2	64 × 54
Standard density – scaled numbers	8 per side	228	±1.6	77 × 47
Scaled density – scaled numbers	8 per side	158	±1.2	55 × 46

Apparatus and test procedures

All matches were played on a sand-based hockey pitch with a standard field hockey ball and the participants own equipment. After fitting participants with a back vest containing a global positioning system (GPS) unit (Optimeye S5, Catapult Innovations, Melbourne, Australia) sampling at 10 Hz, a common 10-min warm-up was performed. Only one match was played each day with all matches played on the same day of the week one week apart. The games were recorded for analysis with a digital video camera (JVC, model GY-HM100, recording at 25 HZ) positioned on the side of the field and elevated 4 m above the pitch. Pre-determined variables were quantitatively analysed by the primary researcher using Sportscode (Sportstec Limited, Sydney, Australia).

Dependent variables

Each game was coded and analysed for the following performance variables:

Successful pass: The successful attempt of a player to deliver the ball to another teammate.

Unsuccessful pass: The unsuccessful attempt of a player to deliver the ball to another teammate.

Total passes: The total of attempts of a player to deliver the ball to another teammate.

Successful dribble: The successful attempt of a player to move while controlling the ball with the stick.

Unsuccessful dribble: The unsuccessful attempt of a player to move while controlling the ball with the stick.

Total dribbles: The total of attempts of a player to move while controlling the ball with the stick.

Skilled actions The sum of total dribbles and total passes.

Successful actions The sum of successful passes and successful dribbles.

Unsuccessful actions The sum of unsuccessful passes and unsuccessful dribbles.

High pressure: The physical pressure applied by a player on an opponent who receives the ball from a teammate within 1 m of the player.

Medium pressure: The physical pressure applied by a player on an opponent who receives the ball from a teammate between 1 and 5 m from the player.

Low pressure: The physical pressure applied by a player on an opponent who receives the ball from a teammate from more than 5 m from the player.

The activity profiles of the participants, captured by the GPS units, were analysed using Catapult Sprint 5.1 software (Catapult Innovations, Melbourne, Australia). The speed zones for walking (0–3 km/h), jogging (3–8 km/h), running (8–13 km/h), high-speed running (13–18 km/h) and sprinting (>18 km/h) were based on previous research examining the activity profiles of youth players in team sports.^13,14

The positional data from the GPS units of all participants were used to calculate the real density per player. Using Matlab R2014A (MathWorks, Natick, Massachusetts, United States), the GPS coordinates were transformed into X- and Y-coordinates using the bottom left corner of the pitch as the origin (see Figure 1). Real density per player was defined as the total space covered by a team divided by the number of field players (Figure 1). A convex hull method was used to measure the total space covered by a team.¹⁵

Figure 1.

Graphical representation of player positions and real density (dashed line) measurement of both teams. Coordination of x-axis and y-axis with origin in the left-bottom corner of the pitch.

Data analysis

A two-way analysis of variance (ANOVA) with repeated measures (with number of players and density as within-participant factors) was used to determine the effect of the number of players and density on the game performance and activity profile of players. Post hoc comparisons were investigated through the use of t-test with Bonferroni correction. To calculate the effect size, partial eta squared ( $η_{p}^{2}$ ) was used, the descriptive terms were: small effect = 0.01, medium effect = 0.06 and large effect = 0.14.¹⁶ The raw mean differences and 95% confidence intervals were reported for significant main effects. Statistical significance was set at p < 0.05.

Results

All presented results are based on the performance of the 11 players who performed all four experimental conditions (means, standard deviations, significant main and/or interaction effects can be found in Table 2). Only the significant effects are reported for the sake of brevity.

Table 2.

Mean (±s) for all the performance and activity profiles variables in the four different conditions.

	Standard density (228 m²)		Scaled density (158 m²)
	Standard numbers (11 players)	Scaled numbers (8 players)	Standard numbers (11 players)	Scaled numbers (8 players)
Successful pass	6.36 (±3.5)	8.55 (±5.0)	5.46 (±3.8)^a	8.64 (±4.9)
Unsuccessful pass	3.36 (±1.6)	2.46 (±2.2)	3.18 (±1.8)	3.55 (±1.4)
Total passes	9.73 (±4.5)	11 (±4.9)	8.64 (±5.2)	12.18 (±5.3)
Successful dribble	1.45 (±1.4)	2.46 (±3.0)	1.45 (±1.8)	2.46 (±2.2)
Unsuccessful dribble	1.0 (±1.2)	0.73 (±1.1)	1.1 (±1.3)	1.82 (±1.9)
Total dribbles	2.46 (±2.3)	3.36 (±3.1)	2.55 (±2.5)	4.27 (±3.6)
Skilled actions	12.18 (±6.3)	14.36 (±6.7)	11.18 (±6.0)^a	16.46 (±7.2)
Successful actions	7.82 (±4.7)	11.82 (±7.3)	6.91 (±4.3)^a	11.09 (±6.1)
Unsuccessful actions	4.36 (±2.2)	3.18 (±3.1)	4.27 (±2.0)	5.36 (±2.7)
High pressure	1.82 (±1.7)	2.64 (±2.0)	0.91 (±1.0)^a	2.54 (±1.8)
Medium pressure	2.64 (±2.1)	3.45 (±2.9)	2.91 (±2.3)	3.09 (±2.2)
Low pressure	1.64 (±1.7)^a	2.18 (±1.3)	1.55 (±2.0)^a	3.73 (±2.6)
Odometer	2217.9 (±314.0)^b	1906.4 (±243.1)	1814.6 (±319.8)	2176.1 (±335.6)^b
Meters/min	87.89 (±12.5)^b	76.37 (±9.7)	72.12 (±12.7)	86.55 (±13.3)^b
Maximum speed	20.06 (±1.4)	19.76 (±2.2)	19.08 (±1.6)	19.98 (±1.5)
Walking distance (m)	213.4 (±77.6)	203.4 (±53.9)	247.0 (±70.6)	204.6 (±45.1)
Jogging distance (m)	1022.6 (±123.2)^c	970.6 (±72.6)	915.3 (±123.6)	1007.1 (±98.3)
Running distance (m)	607.8 (±197.2)^c	481.3 (±186.5)	443.8 (±176.4)	666.5 (±249.8)^b
High-speed running distance (m)	314.1 (±213.3)^b	217.5 (±90.6)	188.4 (±98.8)	267.2 (±124.4)
Sprinting distance (m)	60.0 (±47.6)^c	33.6 (±27.1)	20.2 (±22.3)	31.1 (±19.0)
Surface area per player (m²)	52.8 (±11.2)	46.1 (±12.8)	69.0 (±14.5)	36.7 (±9.9)

Significantly different from the scaled density – scaled numbers conditions.

Significantly different from scaled density – standard numbers and standard density – scaled numbers conditions.

Significantly different from scaled density – standard numbers condition.

Performance variables

Significant main effects for the manipulation of the number of players were found for the following variables: number of successful passes per player (F_1.10 = 4.75, p = 0,05, $η_{p}^{2}$ = 0.33); scaling led to an increment of 2.68 (95%[0.06, 5.42]), number of skilled actions per player (F_1.10 = 7.47, p = 0.02, $η_{p}^{2}$ = 0.43); scaling led to an increment of 3.73 (95%[0.69, 6.77]), number of successful actions per player (F_1.10 = 6.83, p = 0.03, $η_{p}^{2}$ = 0.41); scaling led to an increment of 3.77 (95%[0.56, 6.99]), number of high pressure moments per player (F_1.10 = 8.64, p = 0,02 $η_{p}^{2}$ = 0.46); scaling led to an increment of 1.23 (95%[0.30, 2.16]) and the number of low pressure moments per player (F_1.10 = 5.17, p = 0,05, $η_{p}^{2}$ = 0.34); scaling led to an increment of 1.36 (95%[0.03, 2.70]). A significant main effect for the manipulation of the density was found for the following variable: number of unsuccessful dribbles (F_1.10 = 5.18, p = 0.05, $η_{p}^{2}$ = 0.34); scaling led to an increment of 0.59 (95%[0.01, 1.17]).

Activity profiles

Significant main effects for the manipulation of the density were found for the following variables: Distance covered in 13–18 km/h zone (F_1.10 = 8.15, p = 0.02, $η_{p}^{2}$ = 0.45); scaling led to a decrement of 38.0 (95%[8.34, 67.66]) and distance covered in the >18 km/h zone (F_1.10 = 9.09, p = 0.01, $η_{p}^{2}$ = 0.48); scaling led to a decrement of 21.1 (95%[5.53, 36.83]). Significant number of players x density interactions was found for the following variables: the total distance covered in meters (F_1.10 = 14.17, p = 0.00, $η_{p}^{2}$ = 0.59) (Figure 2(a)), the distance covered per minute (F_1.10 = 13.29, p = 0.00, $η_{p}^{2}$ = 0.57), the distance covered in the 3–8 km/h zone (F_1.10 = 5.12, p = 0.05, $η_{p}^{2}$ = 0.34) (Figure 2(b)), the distance covered in the 8–13 km/h zone (F_1.10 = 12.80, p = 0.01, $η_{p}^{2}$ = 0.56) (Figure 2(c)) and the distance covered in the 13–18 km/h zone (F_1.10 = 7.34, p = 0.02, $η_{p}^{2}$ = 0.43) (Figure 2(d)).

Figure 2.

Graphical representation of the number of players x density interactions for (a). Total distance covered (odometer) in meters (b). Distance covered in the 3–8 km/h zone (c). Distance covered in the 8–13 km/h zone and (d) distance covered in the 13–18 km/h zone. * = significantly different from the scaled density – standard numbers and standard density – scaled numbers conditions. ^ = significantly different from the scaled density – standard numbers condition.

Positional data

The surface area per player revealed a significant main effects for the manipulation of the number of players (F_1.11999 = 25524.66, p < 0.01, $η_{p}^{2}$ = 0.68); scaling led to an decrement of 19.47 (95%[−19.71, −19.24]), manipulation of density (F_1.11999 = 913.95, p < 0.01, $η_{p}^{2}$ = 0.07); scaling led to a an increment of 3.38 (95%[3.16, 3.60]) as well as a significant number of players × density interaction effect was found for the surface area per player (F_1.11999 = 17971.62, p < 0.01, $η_{p}^{2}$ = 0.60),

No significant main and/or interaction effects were found for the following variables: the amount of unsuccessful passes, the amount of total passes, the amount of successful dribbles, the amount of long dribbles, the amount of total dribbles, the amount of unsuccessful actions, the amount of medium (1–5 m) pressure moments, maximum speed and the distance covered in the 0–3 km/h zone.

Discussion

The aim of this study was to examine the impact of manipulating the constraints of density and playing numbers on the match performance in youth field hockey. Results demonstrated that the reduction of the number of players led to an increase in the number of successful passes, skilled actions, successful actions, and high and low pressure moments. Consistent with previous research in basketball³ and soccer,² playing field hockey with 8 players per side, as opposed to the standard 11 players per side, seems to be beneficial in creating additional opportunities for the execution of key field hockey skills. Such conditions provide children with more opportunities to attune to key affordances to freeze information-movement couplings and create stable movement patterns.¹⁷ Lowering the number of players also increased the amount of “high pressure” moments. Although it was expected that lowering the number of players would provide players with more time and space to make their next decision, it seems that playing field hockey with fewer players makes it easier for young children to mark their own opponent, as there are less ‘free’ opponents, and apply more pressure on them. Such pressure moments subsequently resulted in less time for the players to make decisions and perform successful actions (i.e. a dribble or a pass). This time constraint likely forced players to focus on the more specific perceptual information of their teammates and/or opponents in order to guide their movements and perform a successful action.¹⁸ The increase in high pressure moments coupled with the increase in successful actions when playing numbers are reduced, suggests this constraint successfully forces youth players to focus on key information from the environment to guide their movement and positively shape their decision making. A learning environment where children make fewer errors is often associated with an increase in task engagement¹⁹ and an implicit way of learning²⁰ and suggests an avenue for future investigation.

When the individual playing area is reduced, density is increased and vice versa. Results demonstrated an increase in unsuccessful dribbles when density was scaled. This result is consistent with previous research which demonstrated an increase in the amount of dribbles after scaling player density; however, this work didn’t distinguish between successful or unsuccessful actions.⁶ Interestingly, it seems that density doesn’t have a strong influence on the skill performance of participants relative to the number of players. This is likely due to the somewhat artificial nature of the measurement of density which was defined by the relative amount of space available per player. However, this calculation does not consider whether this space is actually used. Comparing the real density per player in the four different playing conditions highlighted that it’s not the playing area but the number of players that is most influential. Manipulating the number of players had the biggest influence on the real density per player; lowering the number of players per side decreased the real density per player by almost 20 m². On the other hand, the density manipulation leads to an increase in the real density per player of about 3 m². From these results, it is clear that not the potential available space dictates the interpersonal distance between players in field hockey, it could potentially be that players interact with each other corresponding to their action capabilities.^21,22 It seems that the affordances in these constrained environments are more influenced by the action capabilities of players than the artificial measurement of personal playing area. For example, the space between or behind two defenders could be accessible to dribble or pass through for one player while another player won’t have the affordances to be able to deal with that situation. This emphasizes the need to constrain learning environments based on the action capabilities of players of a particular age group. The results of this study are not consistent with previous research in soccer that showed that the real density per players significantly decreased when the playing area was reduced.²³ A possible explanation for the difference between soccer and field hockey could be due to the fact that the characteristics of the ball in combination with the action capabilities of the players influences the affordances which makes it easier to use the potential space in soccer than in field hockey. The combination between these constraints also seems to influence the physical demands of youth field hockey players. Scaling the density led to a decrease in high-intensity running (13–18 km/h) and sprinting (>18 km/h), which is consistent with previous research.^4,6 However, both these studies found that the total distance covered was reported to be significantly higher when the individual player area was higher, which was not the case in the current study. Again, the combination between action capabilities and ball characteristics seems to be of influence. It has to be highlighted that the differences in affordances due to various sport-specific task dynamics make it hard to compare the results of manipulating task constraints between different sports. This emphasizes that coaches and trainers should understand the sport-specific task dynamics to manipulate certain constraints to promote specific behaviour.

Furthermore, results of the physical demands show interaction effects for the total distance covered, meters/minute covered, jogging distance (3–8 km/h), running distance (8–13 km/h) as well as for the high-intensity running distance (13–18 km/h). Further exploration of meters/min, which is often used as an indicator of intensity of a training or game,²⁴ revealed that the highest values were found in the standard numbers – standard density and scaled numbers – scaled density conditions, that represented the full-field game and half-field game. It seems that manipulating the number of players or the density made it easier for players to be able to receive a pass from their teammates, while manipulating both constraints forced the players to be more active in order to be able to receive a pass from their teammates.

It can be concluded that match performance of young field hockey players is strongly influenced by the interaction between task constraints and their specific action capabilities. This would indicate that the affordances during game-play are predominantly action-scaled affordances.²² This is of importance as the action capabilities of youth players change over time due to physical growth as well as training, and therefore the competition environment should be scaled accordingly to promote appropriate sport-specific technical and decision-making skills.¹⁸ Future research should aim to find the most influential action capabilities in the guidance of affordances in different sports to create appropriate competition environments.

In summary, the results provide an insight into the effect of manipulating task constraints in skilled junior field hockey and highlight the potential of the constraints-led approach to theoretical describe a change in playing behaviour due to the manipulation of several task constraints. The results also highlight the importance of coaches and practitioners having a clear understanding of the impact of the constraints they choose to manipulate. In particular, it is critical to understand how various constraints interact with each other and in turn establish specific information-movement couplings to enhance skill performance and acquisition.¹² Further research should focus on the different manipulation of constraints and its influence on the development of skill acquisition in sports.

Footnotes

Acknowledgements

The authors would like to thank the coaches, staff and players of the Tigers team for their participation in this study. The authors would also like to thank Cas van Niel and Joel de Water who assisted with the data-analysis in the preparation of this manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Buszard T, Reid M, Masters R, et al. (2016) Scaling the equipment and play area in children’s sport to improve motor skill acquisition: A systematic review. Sports Med 46: 829–843.

Dellal

Chamari

Owen

et al.

Influence of technical instructions on the physiological and physical demands of small-sided soccer games. Eur J Sport Sci 2011; 44: 341–346.

Klusemann

Pyne

Foster

et al.

Optimizing technical skills and physical loading in small-sided basketball games. J. Sports Sci 2012; 30: 1463–1471.

Davies

Young

Farrow

et al.

Comparison of agility demands of small-sided games in elite Australian Football. Int J Sports Physiol Perform 2013; 8: 139–147.

Lupo

Tessitore

Cortis

et al.

A physiological, time-motion, and technical comparison of youth water polo and Acquagoal. J Sports Sci 2009; 27: 823–831.

Casamichana

Castellano

. Time-motion, heart rate, perceptual and motor behaviour demands in small-sides soccer games: effects of pitch size. J Sports Sci 2010; 28: 1615–1623.

Gabet

Abernethy

Jenkins

. Influence of field size on the physiological and skill demands of small-sided games in junior and senior rugby league players. J Strength Cond Res 2012; 26: 487–491.

Newell

Constraints on the development of coordination. In: Wade

Whiting

HTA

(eds). Motor development in children: aspects of coordination and control, Dordrecht: Martinus Nijhoff, 1986, pp. 341–360.

Newell

Broderick

Deutsch

et al.

Task goals and change in dynamical degrees of freedom with motor learning. J Exp Psychol Hum Percept Perf 2003; 29: 379–387.

10.

Davids

The constraints-based approach to motor learning. Implications for a non-linear pedagogy in sport and physical education. In: Renshaw

Davids

Savelsbergh

GJP

(eds). Motor learning in practice. A constraints-led approach, London: Routledge, 2010, pp. 3–16.

11.

Gibson

. The ecological approach to visual perception, Hillsdale, NJ: Lawrence Erlbaum Associates, 1979.

12.

Jacobs

Michaels

. On the apparent paradox of learning and realism. Ecol Psychol 2002; 14: 127–139.

13.

Castagna

D’Ottavia

Abt

. Activity profile of young soccer player during actual match play. J Strength Cond Res 2003; 17: 775–780.

14.

Castagna

Manzi

Impellizzeri

et al.

Relationship between endurance field tests and match performance in young soccer players. J Strenght Cond Res 2010; 24: 227–3233.

15.

Frencken

Lemmink

KAPM

Delleman

et al.

Oscillations of centroid position and surface area of soccer teams in small-sided games. Eur J Sport Sci 2011; 11: 215–113.

16.

Cohen

. Statistical power analysis for the behavioural sciences, 2nd ed. Hillsdale, NJ: Erlbaum, 1988.

17.

Savelsbergh

GJP

van der Kamp

. Information in learning to Co-ordinate and control movements: is there a need for specificity of practice? Int J Sports Psychol 2000; 31: 467–484.

18.

Renshaw

Chow

Davids

et al.

A constraints-led approach perspective to understanding skill acquisition and game play: a basis for integration of motor learning theory and physical education praxis. Phys Educ Sport Pedagog 2010; 15: 117–137.

19.

Farrow

Reid

. The effect of equipment scaling on the skill acquisition of beginning tennis players. J Sports Sci 2010; 28: 723–732.

20.

Maxwell

Masters

RSW

Kerr

et al.

The implicit benefit of learning without errors. Q J Exp Psych A 2001; 54: 1049–1068.

21.

Dicks

Davids

Button

. Individual differences in the visual control of intercepting a penalty kick in association football. Hum Mov Sci 2010; 29: 401–411.

22.

Fajen

Riley

Turvey

. Informational, affordances, and the control of action in sport. Int J Sport Psyhol 2008; 40: 79–107.

23.

Silva

Duarte

Sampaio

et al.

Field dimension and skill level constrain team tactical behaviours in small-sided and conditioned games in football. J Sports Sci 2014; 32: 1888–1896.

24.

Aughey

. Applications of GPS technologies to field sports. Int J Sports Physiol Perform 2011; 6: 295–310.